The idea of the algorithm of extrapolating the nonstationary hydrometeorological series using an optimized empirical model

The problem of extrapolating the nonstationary time series from data of its observations in the past is considered. General requirements to the desired solution are formulated. A number of new elements of the algorithm of problem solving are suggested, which have potential to obtain the more accurate and reliable forecasts of actual hydrometeorological series under climate change. The principal steps in determining the optimal predictive empirical model for the series and constructing, based on it, approximate probability predictions of the values of variables under investigation are discussed.